© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-2B
    4. Residual Stream - 16k
    5. 22-GEMMASCOPE-RES-16K
    6. 3107
    Prev
    Next
    INDEX
    Explanations

    incidents of violence and physical aggression

    oai_token-act-pair · gpt-4o-miniTriggered by @bot

    Reports of physical altercations and violence

    np_acts-logits-general · gemini-2.0-flash

    physical violence and assault

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 22-gemmascope-res-16k
    Configuration
    google/gemma-scope-2b-pt-res/layer_22/width_16k/average_l0_72
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.22.hook_resid_post
    Hook Layer
    22
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     فريبيس
    -0.55
    новништво
    -0.53
     biais
    -0.52
     optimis
    -0.52
    AccessorTable
    -0.51
     Optim
    -0.49
     percolation
    -0.48
    torie
    -0.48
    ắc
    -0.47
     dries
    -0.47
    POSITIVE LOGITS
     violence
    1.39
     violent
    1.26
     brawl
    1.24
     aggression
    1.23
    Violence
    1.21
     fists
    1.21
     punches
    1.20
     Violence
    1.15
    violence
    1.14
     fights
    1.13
    Activations Density 0.477%

    No Known Activations